Large deviation inequalities for the nonlinear unbalanced urn model

Abstract

In the present paper, we consider the two-color nonlinear unbalanced urn model, under a drawing rule reinforced by an ${\mathbb{R}}^{+}$-valued concave function and an unbalanced replacement matrix. The large deviation inequalities for the nonlinear unbalanced urn model are established by using the stochastic approximation theory. As an auxiliary theory, we give a specific large deviation inequality for a general stochastic approximation algorithm.